The Agent Skills Directory

[DATA_EXFILTRATION]: The skill instructs the agent to read sensitive configuration files including root .env and apps/tanstack-app/.dev.vars to verify the presence of the BETTER_AUTH_SECRET variable during the template detection phase.
[INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided strings for product names, pitches, and call-to-action labels and interpolates them directly into project source code and i18n locale files.
Ingestion points: User-provided inputs for 'Product name', 'One-line pitch', and 'Primary CTA label' defined in Step 3 of SKILL.md.
Boundary markers: There are no explicit instructions or delimiters used to ensure the agent ignores potential instructions embedded within these user-supplied strings during interpolation.
Capability inventory: The skill exercises extensive file-write capabilities across the project, modifying .tsx, .vue, and .ts files.
Sanitization: The process lacks mention of sanitization or validation of the user-provided content before it is written to the filesystem.
[COMMAND_EXECUTION]: The instructions direct the agent to handle port assignments and programmatically set environment variables such as APP_BASE_URL and BETTER_AUTH_URL within the local environment to ensure the application starts correctly during the setup and verification steps.

tinyship-recipe